PrimeTime Timing is a full service sports technology company providing customized solutions to active-sport events. Founded in 2003, PrimeTime began with the goal of applying the latest in sports technology to all levels of events. PrimeTime works with over 500 events annually providing a range of services, including custom software scoring solutions solutions, custom graphic packages for broadcast and web productions, online registration, and video display services. PrimeTime’s event resume includes events such as the Prefontaine Classic, NCAA Track & Field National Championships, NCAA Cross Country Championships, SEC Indoor & Outdoor Track & Field Championships, Big 10 Outdoor Track Championships, LifeTime Fitness Triathlon, Nautica Malibu Triathlon, American Birkebeiner Nordic Ski Race. In addition to the headquarters office in Milwaukee, Wisconsin, PrimeTime has satellite offices in Wisconsin Rapids, WI, Menomonie, WI, and Austin, TX.
So the next thing we need is a collection of images to compare our unknown images too. We need a training set. This set could be used to train any sort of AI technique we are going to use (neural network etc…). Using appropriate data for training can make or break your results. The better the dataset the more likely you are to succeed. Since we are targeting a specific CAPTCHA, and since we already can extract the images from the CAPTCHA itself why not use those images as our training set? That’s what I did. I downloaded many of the generated CAPTCHA images and had my program break them up into letters. I then organised my images into a collection (corpus). After a few iterations I had at least one example of every letter and number that the CAPTCHA produced. Adding more examples to my corpus would increase accuracy but what I had was enough for proof of concept. The code to produce the images is below.
About the TPC
Who We Are
Benchmarks Enterprise BMs TPC-C
Express BMs TPCx-BB
Common Specifications TPC-Pricing
Obsolete BMs TPC-A
Join the TPC
TPCTC What's New 11/20/2017 Cisco publishes the first TPCx-IoT benchmark result.